The scandal isn’t that TV detector vans are going to intercept your Wi-Fi. The scandal is that you still believe in TV detector vans

No, the BBC can’t drive up your street and sense that you’re using iPlayer. And it probably never could tell if you were watching TV

You’d better pay your TV licence fee. Day and night, the BBC’s detector vans roam the streets seeking out unlicensed viewers. You could be next!

We haven’t heard quite so much about these four-wheeled agents of the surveillance state in recent years, but they’re back in the news because the Daily Telegraph ran a front-page story on plans to update them to detect people watching iPlayer, the BBC’s online live and catch-up viewing service, which is available free of charge within the UK. Previously, a licence was required to watch live programmes on iPlayer, in just the same way as if you watched them using a TV aerial, but not if you watched them later. From September, a licence will be required to watch any TV content on iPlayer.

‘The Daily Telegraph can disclose,’ said the Daily Telegraph, ‘that from next month, the BBC vans will fan out across the country capturing information from private Wi-Fi networks in homes to “sniff out” those who have not paid the licence fee. The corporation has been given legal dispensation to use the new technology, which is typically only available to crime-fighting agencies.’

‘The disclosure will lead to fears about invasion of privacy,’ predicted the paper. No kidding. So rapidly did those concerns flood social media that within a day the BBC was issuing a statement that ‘it is wrong to suggest that our technology involves capturing data from private wi-fi networks’.

This seemed to render moot an emerging debate between computer security experts about how the BBC might detect, from the kerb, that people were accessing iPlayer, along with a parallel debate between lawyers about how it might do so without its staff or contractors going to prison.

Capturing data from private networks would certainly require some kind of legal cover. The Regulation of Investigatory Powers (British Broadcasting Corporation) Order 2001 already permitted TV Licensing, the BBC’s licence enforcement arm, to use powers created in the notorious Regulation of Investigatory Powers Act 2000 (RIPA) to deploy detection equipment. If this was to be amended in the even more notorious Investigatory Powers (IP) bill currently going through parliament, there was no sign of it in existing drafts. But if no data was to be ‘captured’ at all from private networks, what would TV Licensing be requesting permission for? What would the vans be doing?

According to a recent National Audit Office report:

1.37 The BBC’s final detection and enforcement option is its fleet of detection vans. Where the BBC still suspects that an occupier is watching live television but not paying for a licence, it can send a detection van to check whether this is the case. TVL detection vans can identify viewing on a non‐TV device in the same way that they can detect viewing on a television set. BBC staff were able to demonstrate this to my staff in controlled conditions sufficient for us to be confident that they could detect viewing on a range of non‐TV devices.

‘In the same way?’ But how could that be, when terrestrial broadcast and internet protocol (‘IP layer’ — geddit) delivery are completely different technologies?

Of course, this might just be a regrettably vague form of words, meaning merely that just as traditional TV viewing can be detected, so can iPlayer viewing (but perhaps by quite different means). Either way, it raises another question. After it became practical to ‘watch TV’ on various devices (such as PCs with TV tuner cards) other than a ‘TV set’, the licence fee legislation was retconned to refer to watching live broadcasts on any device. But how could the TV licence detector vans use the same techniques to detect the use of those devices as TV sets? What were those techniques?

And, for that matter, how many detector vans are there, and how often are they actually involved in identifying and prosecuting TV licence evaders?

Officially, we don’t know. We don’t know because the BBC doesn’t want to say and the Information Commissioner’s Office says it doesn’t have to.

‘The BBC has stated that the information is exempt from disclosure as it would prejudice the prevention or detection of crime… due to the fact that a person would use the information to evade the licence fee,’ the ICO reported in a 2008 ruling, accepting the BBC’s argument and noting that it ‘relies on the public perception that the vans could be used at any time to catch evaders. This perception has been built up since the first van was launched in 1952.’

Honestly, I can’t imagine any form of words closer to ‘We made all this shit up to scare people in a more innocent age, don’t spoil it now,’ short of just blurting it out.

I had a DM conversation today with a national newspaper journalist who was trying to write something up on this. Neither of us had found any primary source who could confirm whether TV detector vans were real or fake. Both of us thought this was a good (though not conclusive) indication that they were fake.

Where are the thousands of alumni of the TV licence detection squad — designers, engineers, drivers — with their stories to tell? Nobody claims this stuff is a state secret. Even the SAS has its memoir hawkers. The only way this is still a secret is that there’s no secret. Hardly anyone has ever been closely involved with ‘TV licence detection vans’ because there aren’t any TV licence detection vans. There were a couple of unconvincing props a while back. That’s it. Not even Fox Mulder would look twice at this file.

Documentary evidence is also loquacious by its absence. In 2011, the BBC admitted, in response to a Freedom of Information Act request from the TV Licensing blog, that it had ‘not, to date, used detection evidence in Court’. Ever. What was the point, then? A BBC lawyer explained:

Under TVL’s current prosecution process the presentation of detection evidence in court is unnecessary. This is because TVL uses detection evidence when applying for search warrants. If, following service of the warrant an individual is found to be evading payment of the TV Licence, then the evidence obtained via the search warrant is used in court, not the detection evidence.

Uh, OK. Evidence that’s good enough to stick in front of a busy magistrate to get a quick warrant, but not good enough to be worth presenting in court later to support your case. Sure. That kind of evidence.

We can see this process in action in a rare documented case from 2011. A post on the TV Licensing blog, unearthed by lawyer Graham Smith, reveals one of the very few legal papers referring to the operation of ‘licence detector vans’, again obtained through an FOIA request to the BBC. This arose from the case of Steve Heather, who claimed — quite honestly, as it turned out — that he never watched broadcast TV, although he did use various TV equipment for other purposes.

Under oath, in support of an application for a search warrant, one Chris Cristophorou from TV Licensing submitted that:

6. On the 31st March 2011 at 18.36 hours the detector van was positioned near the Premises. When the detector camera was pointed at the window of the Premises a positive signal was received indicating a TV receiver was in use receiving a possible broadcast with a confidence factor of 97%.

So many questions.

First of all, the ‘confidence factor of 97%’. They’re 97% confident of what? As an arts graduate occasionally called upon to struggle with statistics, I’m no authority on scientific method, but I know hand-waving when I see it. The basic assertion here is that the light coming from the suspect’s screen matched a programme being broadcast at that time, proving that they were watching that programme, which required a TV licence.

But does ‘97% confidence factor’ mean that 97% of the data captured matched the data for the simultaneously broadcast programme? Surely not so simple, and we’d still need to decide what to think about the 3%. Or does it mean that whatever the nature and degree of similarity identified, this was calculated by some method to represent a 97% likelihood that what was being watched was indeed this programme? More likely, but introduces unknown variables. Is the method valid? Has it been published? (Of course not. The BBC doesn’t talk about how detection works, ‘for obvious reasons’.)

An explanation of the detection system is also provided, but it’s not, shall we say, massively helpful:

5. A television display generates light at specific frequencies. Some of that light escapes through windows usually after being reflected from one or more walls in the room in which the television is situated. The optical detector in the detector van uses a large lens to collect that light and focus it on to an especially sensitive device, which converts fluctuating light signals into electrical signals, which can be electronically analysed. If a receiver is being used to watch broadcast programmes then a positive reading is returned. The device gives a confidence factor in percentage terms, which is determined by the strength of the signal received by the detection equipment and confirms whether or not the source of the signal is a “possible broadcast”.

This reminds me of the worst kind of patent application. Why are we told that the light is ‘at specific frequencies’? All light is at specific frequencies. A colour TV picture, by design, includes pretty much all the frequencies visible to humans, which isn’t very specific. The optical detector uses a ‘large lens’? Not all large lenses are fast and not all fast lenses are large, so meh.

It focuses the light ‘on to an especially sensitive device’ — oh yawn, this is going to be a CCD or CMOS image sensor that you could get for a fiver in Maplin’s; it’s more sensitive than the one in the packet on its left and less sensitive than the one on the right. It ‘converts fluctuating light signals into electrical signals, which can be electronically analysed’ — yes, that’s what an image sensor does, congratulations again on not using a house brick.

And thus… ‘the device gives a confidence factor’. Yay! Wait, was there a bit missing there? You know, where it actually DID THE THING?

To summarise:

  1. Point camera in general direction of TV
  2. A little app that somebody wrote, of which we haven’t seen the source code, compares the aggregate RGB levels (we’re only going to be dealing with the screen as a whole here, not reconstructing an image pixel by pixel, because this is not CSI: Miami) with those of a current live TV broadcast, at some unspecified time interval
  3. The app uses a formula we haven’t seen to give you a percentage confidence that something or other is the case (that the data matches? That the data matches and it wouldn’t match any other programme’s data? That the data matches and it wouldn’t match some other programme’s data and the data isn’t erroneous?)
  4. You take this number to a judge and they go ‘97%, well, sure’*

*This is apparently what happened, because TV Licensing got the warrant. As a result, they were able to enter Steve Heather’s home (accompanied by police) against his wishes, ‘test’ his TV set by turning it on, and briefly see something that they claimed was a live BBC One broadcast but the defence claimed was the Sky EPG that Steve used to select radio programmes. The end result was that TV Licensing lost the case and Steve didn’t have to buy a TV licence to not watch TV.

I wonder why they needed to point a ‘detector camera’ at the window at all? Why could the surveillance person not just look through the window with their eyes, see Antiques Roadshow on the TV and make a sworn statement to that effect? Was the suspect’s television screen invisible to the naked eye from the window through which its reflected light could be detected? It seems like bad luck if so. I could tell you what most people in my street were watching at any given time just by walking around.

I probably wouldn’t think it was worth designing a whole system specifically for the couple of houses with their TVs in exactly the wrong place. Especially if I already had a database of all the people who had TV licences, narrowing down the list of potential infringers by about 95%. (We’ll return to this later.)

Let’s imagine any of this made any sense. Is such a technology plausible? One anonymous commenter thought so:

All they need in the van is a computer to ‘watch’ TV and compare the average light level of the screen at each frame (possibly as three different levels (red, green and blue) for extra confidence) and compare that with the light fluctuation coming out of the windows, chuck in some VERY simple pattern matching software that any first year computer science student could write and you’re good to go.

I agree: this is a fun one-day hackathon project, not a Large Hadron Collider. But in practice it founders on the obsolete assumption that we know what’s ‘on TV’ at a given moment.

A TV licence is required to watch any content ‘as it’s being broadcast’. If you have a TV set (or any other device capable of displaying a TV picture) and only use it to watch ITV, you still need a TV licence. If you have a TV set and only use it to watch Sky cable (for which you’ve paid Sky), you still need a TV licence. If you have a TV set and only use it to watch programmes broadcast within North Korea, which you’ve somehow managed to intercept, then as long as you watch them ‘live’, not from a recording, you still need a TV licence, as far as we can tell from the legislation (case law links welcome).

Clearly, a computer ‘in the van’ wouldn’t be able to watch more than a fraction of live broadcasts, presumably using a portable TV aerial. (One imagines the operator exasperatedly adjusting it.) More plausibly, TV Licensing might have a facility on premises somewhere to do the watching, but this still doesn’t make practical sense: it would need equipment, connections, decoders and permissions to ‘watch’ a range of cable, satellite and terrestrial channels. The results would have to be recorded on a rolling basis, presumably in some sort of proprietary minimised encoded format, for matching against data from the vans. This is starting to look more of a moonshot.

I suppose the BBC could choose to match only live BBC programmes, using its own data. But it would seem bizarre to go to all the trouble of staking out a suspect’s home with specialist equipment, only to test for a tiny subset of the possible evidence that would incriminate them.

‘Got a result for me yet, Sponson?’ ‘Nah boss, the muppet’s watching QVC again. I can hear him buying a vegetable slicer but we can’t get eyes on.’

Let’s say this really was done in the case of Steve Heather (who really did never watch TV with the curtains open, because, as a court ultimately accepted, he never watched TV at all.) It should still give very little pause to anyone who is planning to watch TV without a licence, because there’s still no reason to think more than one van exists, if that. All you’d need is a laptop connected to a positionable webcam pointing out of the window. It could even be stuffed in the back of a car when required.

Of course, that’s just my speculation. If TV Licensing really did only have a small number of basic contraptions that it could use ad hoc as cover for legal action when necessary, it obviously wouldn’t be able to criss-cross the country day in, day out, in a relentless search for licence fee evaders.

If it was only going to be able to pick on a few intransigent non-payers, though, Steve Heather would have been a handy choice, what with him living in Kent, no more than an hour’s drive from the Maidstone headquarters of TV Licensing’s contractor and Chris Christophorou’s employer, Capita Business Services.

Here’s the thing about this whole business of matching the light emitted from a suspect’s screen to a simultaneous broadcast. To a large extent, it makes sense — more sense than some other entertaining ways in which people have explained TV detection (although some sort of oscilloscope test for radio reception might actually have worked in the early days). It would be a method of confirming, to some degree of confidence, that someone was watching, on a screen obliquely or indirectly visible from the street, a particular programme, independently of what type of device they were using, how they were getting the signal, and what screen technology (CRT, LCD, plasma and so on) was displaying the picture.

But it would be useful only in a tiny number of cases. And this is the central flaw in the whole tale of a fleet of vans routinely trawling the country for licence fee evaders.

The BBC’s wall of silence was accidentally breached in 2015, when it released, as part of an FOIA response to WhatDoTheyKnow, a performance report from Capita that recorded just 116 detection requests across the UK in the whole preceding year. If detection vans really had been ‘fanning out’ across the country, they’d been driving around either conducting large-scale illegal surveillance operations or doing absolutely nothing except at those 116 addresses — assuming all 116 requests were approved by the relevant senior BBC officials.

Almost everyone (95% of UK households) watches TV. Everyone I know also uses iPlayer. There’s simply no need — and hasn’t been for most of the time the BBC existed — to go around detecting who’s using these services. It’s a working assumption that everyone does. If someone doesn’t have a TV licence, TV Licensing can write to them and threaten to use its powers to send inspectors round to check. This is something that certainly does happen on a large scale.

People don’t have to let the inspectors in (and, like Steve Heather, they can even pre-empt the knock by formally ‘withdrawing the implied right of access’ that, under common law, permits the licence inspectors to mosey up their garden path). But unless they successfully convince TV Licensing that they never watch live TV, they face further action leading ultimately to criminal prosecution and a fine of up to £1,000 plus costs or even imprisonment. This happens to as many as 200,000 people a year, often with distressing consequences.

Compare the 200,000 prosecutions to the 116 RIPA requests for detection. Does that suggest actual, functioning detector vans are a significant element of TV Licensing’s deterrence and enforcement strategy? Or that they’re smoke and mirrors?

It’s time to put this charade to bed. When you have organisations like the ICO and the National Audit Office playing along — exercising, I can only surmise, a kind of Orwellian doublethink in not being overtly in on the joke, yet discreetly avoiding the obvious awkward questions — and national newspapers apparently being briefed with fictional accounts of the deployment of non-existent resources for nonsensical purposes, the public is being misled in a way that can’t be squared with a modern conception of transparent and accountable democracy.

And when a front page story has the BBC exercising brand new powers to intercept private electronic communications during the controversial passage of a bill, personally associated with the current Prime Minister, that attempts to navigate what it acknowledges to be the minefield of such surveillance, it’s painfully clear that the patrician fairy stories of the 1950s just won’t wash in 2016.